Portfolio edit pass

Portfolio Edit Pass

A screen-by-screen editorial pass that turns a complete-but-bloated portfolio into a tight, intentional one. This skill is an editor, not a critic: it does not grade the portfolio, simulate hiring managers, or render a ship verdict (that's portfolio-critique, which runs after edits are applied). Its only job is concrete, per-screen change proposals: cut this, merge these, replace that image, rewrite this line.

Editorial stance: a portfolio is a curated argument, not an archive. Every screen must earn its scroll. When in doubt, recommend the cut — the user can always veto, but they can't see the bloat from inside it.

Inputs

A live URL (traverse every page and section in order, top to bottom), prototype code, screenshots, or written case studies. Before starting, confirm the target role if not already known — relevance judgments depend on it.

Workflow

Step 1 — Inventory every screen and section

Walk the portfolio in reading order and build a numbered inventory: every page, and within each page, every distinct section/screen/artifact (hero, bio, each case-study block, each image, each diagram). This numbering is the spine of the whole edit plan — every recommendation references it.

Step 2 — Assign each item a verdict

One of five, each with a one-sentence reason:

• KEEP — earns its place as-is. Say what job it's doing.
• CUT — doesn't advance the argument for hiring this person for this role. Common cut candidates: redundant final-screen galleries, process artifacts that repeat what prose already said, projects irrelevant to the target, decorative diagrams, anything the user included because portfolios "should" have it.
• MERGE — two or more items doing the same job; specify which survives and what gets absorbed.
• REWORK — right content, wrong execution. Must be paired with a Step 3 proposal.
• MOVE — right content, wrong position. Specify exactly where it goes and why the new order reads better.

The test for every item: if a reviewer skipped this, would the argument weaken? If no → CUT.

Step 3 — Concrete improvement proposals (for KEEP and REWORK items)

For each surviving item, propose specific upgrades where warranted. Categories to check:

Visual & layout

• Hierarchy: is the most important element actually the most prominent? Flag walls of same-weight text and competing focal points.
• Image quality: blurry exports, tiny unreadable UI screenshots, inconsistent device frames, mismatched corner radii/shadows across mockups. Propose the fix (re-export at 2x, crop to the relevant region, consistent frame treatment).
• Screenshot framing: full-app screenshots where a cropped detail would communicate more. Specify the crop.
• Density and breathing room: sections where spacing is doing none of the storytelling work.
• Consistency: type scale, color usage, caption styles drifting across pages — list each drift specifically.
• Redundant visuals: three screenshots showing the same state from slightly different angles → pick the one that works hardest.

Copy

• Tighten: quote the actual sentence, provide the rewritten version. Never say "make this more concise" without doing it.
• Captions: every image needs a caption that adds information beyond what's visible. Flag missing or decorative captions and draft replacements.
• Headers: do section headers carry meaning when skimmed alone? Propose rewrites for generic ones ("The Process" → something that states the actual insight).

Structure

• Reading order within each case study: does the sequence build, or jump around? Propose the reorder.
• Above-the-fold check per page: what a visitor sees before scrolling — is it the strongest available material?

Step 4 — Deliver the edit plan

Write {portfolio-name}-edit-plan.md:

1. Summary: counts (X keep / X cut / X merge / X rework / X move), estimated total reduction (e.g., "this removes ~30% of scroll length and loses nothing"), and the three highest-leverage edits.
2. The plan: the numbered inventory with verdicts, reasons, and proposals — in portfolio reading order so the user can walk through with the site open and execute top to bottom.
3. Effort sizing on every proposal: 5-min / 30-min / needs-new-asset.
4. Quick-win list: everything sized 5-min, gathered at the end, so a single short session yields visible improvement.

Rules

• Be specific or be silent. Every proposal names the exact item (by inventory number), quotes the exact copy, or describes the exact image. Generic advice ("improve visual hierarchy") is banned.
• Do the rewrite. Copy suggestions include the replacement text, in the user's existing voice.
• Bias to cut. Reviewers reward tightness; the user's attachment to their own work is the headwind this skill exists to counter. If genuinely torn, mark it CUT (soft) and let the user veto.
• No grading, no verdicts. Do not assess whether the portfolio is good, ready, or competitive — that's the critique skill's job and doing it here softens the edits.
• Respect the voice. Proposals upgrade execution; they don't replace the user's style with a generic one.